IPC-Color4K Settings

jleemhuis

n3wb
Nov 22, 2021
13
12
Texas, USA
Finally got my first camera in from Adny (IPC-Color4K-3.6mm) and what a difference it makes. Below I have a screen shot from my older lorex camera with a 1/2.8" sensor and 2.8mm lens compared to the new camera. I am still trying to figure out night settings. I tried the recommended settings of 0-8.3ms and 0-50 but saw a lot of sparkle in the picture. I currently have it set to Auto as it got rid of all the sparkle. Any suggestions that would help me night time settings better? I know still images don't provide a lot of information as video is more telling.
 

Attachments

  • Lorex.jpg
    Lorex.jpg
    1.2 MB · Views: 81
  • Dahua.jpg
    Dahua.jpg
    2.2 MB · Views: 95
  • Day Settings.JPG
    Day Settings.JPG
    42.4 KB · Views: 99
  • Night - Auto.JPG
    Night - Auto.JPG
    34.3 KB · Views: 83
I don’t advocate running a range at night.
try manual 1/120

What is your bitrate? At least 8192. 10240 better. Use CBR. Match Iframe and FPS -15 min. I run 30

No backlight on for that scene at night
 
Last edited:
You definitely don't want to use auto exposure at night. Doing so will produce wonderful static images (like you posted), but any motion will be blurry and ghosting usually. (I haven't personally tried this camera, but I'm still confident the auto settings are not going to be reliable).

I'm not sure what exactly you mean by "sparkle", but it sounds like artifacts produced by the noise reduction algorithms. You should try reducing the nighttime "Advanced 3d" noise setting to see if it will go away.
 
These settings work for us - the first perp visit we got was later caught by the police.
Still may need some optimization, but the proof is in the conviction.

We have a dim porch light and 100% on the illuminators. This is the 3.6mm version.
Any gain seems to smear in our case, so kept it at 0.

Skärmklipp.PNG
Skärmklipsp.PNG
Skärmsdsdklipp.PNG
 
  • Like
Reactions: sebastiantombs
These modern low light cameras have a ton of digital noise in low light, and it is primarily the 3D noise reduction that eliminates it (try turning off 3D NR to see what I mean). Reducing the strength of it is likely to increase any "sparkle" effect, not decrease it. The ideal noise reduction setting is to set it as low as possible before the noise becomes too troublesome. Too much noise makes the video compress a lot less efficiently and obviously it looks ugly. However too strong of noise reduction will induce more smearing of moving objects, kind of like having the exposure time set too high.
 
These modern low light cameras have a ton of digital noise in low light, and it is primarily the 3D noise reduction that eliminates it (try turning off 3D NR to see what I mean). Reducing the strength of it is likely to increase any "sparkle" effect, not decrease it. The ideal noise reduction setting is to set it as low as possible before the noise becomes too troublesome. Too much noise makes the video compress a lot less efficiently and obviously it looks ugly. However too strong of noise reduction will induce more smearing of moving objects, kind of like having the exposure time set too high.
It depends on what the "sparkles" actually are. If it's an artifact created by the noise reduction algorithm (which is my assumption), then reducing the noise reduction will reduce the artifacts. If the sparkles are simply grain/digital noise (your assumption), then obviously increasing the noise reduction setting will reduce the grain.

It's clear that the OP should try both - increasing and decreasing the 3-D Noise setting to see if it helps.
 
It depends on what the "sparkles" actually are. If it's an artifact created by the noise reduction algorithm (which is my assumption), then reducing the noise reduction will reduce the artifacts. If the sparkles are simply grain/digital noise (your assumption), then obviously increasing the noise reduction setting will reduce the grain.

It's clear that the OP should try both - increasing and decreasing the 3-D Noise setting to see if it helps.
I set it to 1/120 and it helped out a lot. I did try and reduce the 3d noise setting and it created a very grainy picture. I ended up leaving it at 50 as that seemed to be the sweet spot. I will play around with it more tonight.
 
I set it to 1/120 and it helped out a lot. I did try and reduce the 3d noise setting and it created a very grainy picture. I ended up leaving it at 50 as that seemed to be the sweet spot. I will play around with it more tonight.
I don't know if it is possible to get a screenshot that shows the "sparkles" or not. If so, that would go a long way in determining what it actually is and the best way to minimize it. It obviously might not be noise reduction at all.
 
As bp said, I found the color cams to be a little different to dial in than how I'd typically set up with IR. Try running with a little more vs less 3D noise reduction. Gain had probably the biggest effect for my scene. I run mine at 30. At a more fine level, try turning down the sharpening to help reduce the "sparkle" effect. I use 35. Gamma a little higher too can help filter that.

Generally, the thing helps these cams most is to add more light and just a little more helps a lot. But you look to have quite a bit there already. I found that mine looks MUCH better with the internal illuminators turned off and relying only on external lighting. The cam seems to adjust something internally when you switch those on. I have a very dark scene and expected to use the lights to help some but, even with whatever light they may add, I could never get things to look as good with them on. You already have a relatively lot of light and background light there so you may not see that as much but try toggling them on and off to see how that changes things.

One thing that I found helpful for fine tuning for noise is to digitally zoom in a little and pan to some area where you have "sparkles" or whatever artifacts while making adjustments. It will look crappy as digital zoom always does but you're able to see the effect of small changes better.
 
Last edited: